A Lifeless Earth Model Could Stay Habitable—and That Changes the Search for Life

Imagine a twin Earth where life never began. It has oceans, continents, plate tectonics, and a Sun that slowly brightens—but no photosynthetic oxygen, no methane-producing microbes, no forests. Would it still maintain a climate fit for liquid water? A new preprint by Samantha Gilbert-Janizek of the University of Washington and colleagues builds exactly that planet in a computer. The model runs from 50 million years after the magma-ocean phase to 5 billion years, and the authors compare its state at 4.5 billion years to the pre-industrial Earth, the period before the industrial revolution around 1800.

The question is not idle. It goes to the heart of the Gaia hypothesis, the idea that life helps regulate Earth's environment. In its weak form, Gaia says interactions between living and nonliving processes can have some stabilizing effect. In its strong form, life is the most significant process stabilizing Earth's climate. The paper's result bears on the strong version: it suggests that an Earth-like planet can maintain habitable surface temperatures without biological feedbacks, so a wet, temperate world need not be inhabited.

A whole-planet Earth without life

The authors call their approach a whole-planet model: a coupled core-mantle-crust-climate model that co-evolves the deep interior, surface, atmosphere, and evolving Sun. The authors call this an abiotic—life-free—Earth. It starts 50 million years after the magma-ocean phase and runs to 5 billion years. It builds on VPLanet for core-mantle thermal evolution, adds a carbon cycle and a deep water cycle driven by plate tectonics, and uses Clima for wavelength-dependent radiative transfer. A possible basal magma ocean—a molten layer at the base of the mantle—is included as a source of radioactive heating; the paper notes that its existence is still debated.

Carbon cycles among mantle, plate, and surface reservoirs; water cycles between mantle and surface, with surface water partitioned between ocean and atmosphere. Volcanic outgassing adds CO2; silicate weathering and seafloor weathering remove it; subduction buries it. The brightening Sun raises incoming radiation. The model then solves for a surface temperature where absorbed sunlight balances outgoing infrared radiation. The result is not a snapshot but a history from 50 million years after the magma-ocean phase to 5 billion years.

The thermostat does not need life

At the 4.5-billion-year comparison point, the abiotic model reaches a pre-industrial-like state. Its global average surface temperature is 286.9 K, matching the measured pre-industrial value of 286.9 ± 0.1 K. Its atmospheric carbon dioxide partial pressure—a way of expressing how much of the atmosphere is CO2—is 28.2 Pa, close to the measured 28.4 ± 0.4 Pa. It retains about one terrestrial ocean of surface water, or about Earth's ocean mass. Across 19 pre-industrial properties, 17 fall within 1σ of measurements, where σ is the stated measurement uncertainty. The remaining two ocean-chemistry values are within 1.4σ and 2.0σ. Ocean pH is the largest mismatch: the model gives 8.28, while pre-industrial estimates are 8.18 ± 0.05, about a 2σ alkaline offset.

The key result is that the model stays habitable without biology. Surface temperatures remain between about 272 K and 290 K through the run. Early in the run, after a brief equilibration period, CO2 rises to roughly 1 bar because weathering is slow and outgassing is high. As the planet warms, weathering accelerates and draws CO2 down. This carbonate-silicate cycle—a long-term carbon thermostat—acts as a negative feedback: warmer conditions tend to remove greenhouse gas, cooler conditions tend to let it build up. In the model, that thermostat operates with only CO2 and water vapor as greenhouse gases, no life required.

A synthetic HWO spectrum as a null hypothesis

The authors then ask what such a world would look like in reflected light—the light a planet bounces back from its star, split by wavelength. They use the PICASO code to generate synthetic spectra from 0.2 to 2.0 micrometers, the range relevant to the future Habitable Worlds Observatory (HWO), a space telescope designed to directly image and characterize Earth-like planets around Sun-like stars.

Compared with a modern cloudy Earth, the abiotic spectrum lacks the fingerprints of life, or biosignatures. It has no oxygen A-band near 0.76 micrometers, no ozone absorption in the near-ultraviolet and visible, and no methane feature near 1.69 micrometers. It still shows water vapor and carbon dioxide. That makes the synthetic spectrum a control sample: a null hypothesis for what a habitable but lifeless planet might look like. The paper is not reporting an observed exoplanet; it is a proof of concept for how model-generated atmospheres can be turned into predicted telescope signals.

Where the model is still thin

The caveats matter. Clouds are simplified. The model paints clouds onto the surface rather than forming them in 3D, and it needed an unrealistically low cloud fraction of 8% to match pre-industrial temperature; Earth's average cloud cover is about 67%. Ocean chemistry is also simplified. The authors tune calcite saturation to match dissolved inorganic carbon, which helps explain the 2σ pH offset. They neglect methane, nitrogen cycling, atmospheric escape, and the magma ocean phase, and the model is one-dimensional, so it cannot capture snowball Earth transitions or regional climate.

The work is a preprint and has not yet completed peer review. It also does not validate the model against the early Earth, where conditions are poorly known. So the result is not proof that any Earth-like planet will remain habitable without life. It is a model result for one Earth-like case: under a plausible set of abiotic processes, a planet can self-regulate for billions of years.

A control sample for the search for life

That is why the lifeless Earth matters. If habitability can persist without biology, then a temperate, ocean-bearing exoplanet is not automatically inhabited. Distinguishing inhabited from uninhabited-but-habitable worlds will require detecting atmospheric gases or surface features that can only be produced by life—for example, oxygen and methane out of equilibrium on a modern-Earth-like world. The abiotic Earth in this model is not a dead end. It is a baseline. Future telescopes will need to distinguish a living world from a world that merely looks alive. This simulation gives them a model-generated baseline to compare against.